VALIDATION AND IMPROVEMENT OF SPPS IN LOW LIQUID LOADING AND ANNULAR FLOW

Transcription

1 VIII-1 VIII. VALIDATION AND IMPROVEMENT OF SPPS IN LOW LIQUID LOADING AND ANNULAR FLOW Introduction Solid particle erosion of piping and fittings is caused as a result of entrained sand particles in the flow impacting the inner surface of pipe and fittings. Erosion is one of the major concerns in the oil and gas industry. Every year the oil and gas industry suffers the loss of several million dollars in the form of loss in production and repair costs. Solid particle erosion occurs due to entrained sand particles in the flow crossing streamlines and impinging on the inner surfaces of pipes, fittings and other components. Certain components that redirect the flow such as pipe elbows, T-joints and other components are more susceptible to erosion damage. In order to minimize the damage due to this phenomenon, investigators at the Erosion/Corrosion Research Center (E/CRC) are working on developing more accurate models to predict sand erosion damage. Prediction of erosion in multiphase flow is a complex problem due to lack of accurate models for calculating particle impact velocities in multiphase flow. The particle impact velocity is affected by the pipe geometry, carrier fluid velocity, flow pattern, and particle size and distribution in the flow. The complexity of erosion prediction increases significantly for two-phase flow in elbows because of complicated flow patterns that occur when both liquid and gas are present in the flow. Among different flow patterns in horizontal and vertical flows, severe erosion occurs when particles are transported in gas dominant systems such as low liquid loading gas and annular flows. Recent experiments have shown that for annular flows erosion does not consistently decreases with an increase in liquid rate for a given gas rate. For fluid transport application in a number of industries, from compact heat exchangers to large diameter deep-water risers in hydrocarbon production, extension of the current knowledge on gas/liquid flows in a wider range of pipe diameter is necessary. In oil and gas industry, with the increase of demand and major discoveries of hydrocarbon fields becoming rarer in the conventional offshore the deeper water exploration is emerging. Risers employed for deep water are normally operated under friction dominant conditions. To minimize pressure losses, risers tend to be of diameters larger than those for which both multiphase flow and erosion research data are available [1]. It has been recognized [2] that gas/liquid flow in such larger diameter pipes is different from that in smaller pipes. In many industrial applications such as heat exchangers, process equipment and oil and gas industry it is important to be able to predict pressure drop, friction and heat transfer

2 VIII-2 EROSION/CORROSION RESEARCH CENTER in various pipe sizes. In the oil and gas industry, because of large transport pipelines, it is important to have a good understanding of multiphase flow in larger diameter pipes. Also, for predicting erosion, it is important to be able to calculate particle impact velocities in larger diameter pipes. Therefore, it is essential to have a better understanding of multiphase flow in various pipe fittings in order to predict erosion. Furthermore, experiments conducted at The University of Tulsa Sand Management Projects (TUSMP) have shown that flow orientations impacts the severity of erosion in annular flows. The Erosion/Corrosion Research Center (E/CRC) and The University of Tulsa Sand Management Projects (TUSMP) have years of experience in erosion prediction and sand detection. The investigators at E/CRC have developed several user friendly mechanistic models to predict metal loss rates. One of the most prominent models is in the form of the Sand Production Pipe Saver (SPPS) computer program developed at the E/CRC to predict material loss caused by sand impacts for different conditions. Erosion data collected from Harwell, DNV as well as The University of Tulsa E/CRC and TUSMP served as an erosion data bank for modeling work. The SPPS extension for multiphase flows was developed based on limited data for 1 and 2 pipe diameters. However, some recent measurements collected at TUSMP for larger diameter pipes indicate that the SPPS model must be validated and improved [3]. The main goal of this research is to collect additional data in multiphase annular and low liquid loading regimes and develop a more accurate extension of the SPPS procedures that more accurately accounts for the physics of multiphase flows for gas dominants flows and the effect of pipe size and pipe orientation on the severity of the amount of erosion. Objective and Approach The primary objective of this research is to investigate erosion behavior for gas dominant multiphase flows. Additional experiments will be conducted in multiphase annular flow and low liquid flow conditions. In addition, using new empirical data that will be gathered at the E/CRC for low liquid loading and annular flow conditions and new developments in erosion calculations for different oilfield materials, the results of this investigation will be applied to validate and improve the E/CRC erosion prediction model (SPPS), thus making SPPS more reliable for predicting erosion in larger diameter pipes and different pipe flow orientations.

3 VIII-3 Background of Gas Dominant Flows Background and Literature Review Low-liquid loading and annular gas-liquid flow conditions are commonly encountered in gas transportation pipelines. They may also occur in other production facilities as gas/condensate production systems where the liquid phase flows partly as a wavy film along the pipe circumference, and partly as entrained droplets in the turbulent gas core. A typical gas/condensate production system contains mainly methane (CH 4 ) and propane in addition to impurities such as ionic formation water, CO 2 and H 2 S. Also, several wells contain sand that can cause erosion. In addition, provided the wall is wetted by corrosive water, the presence of small quantities of sand and CO 2 could result in erosioncorrosion of production equipment. The presence of small quantities of sand particles will normally not alter the global flow conditions significantly, but could, by interaction with the confining walls, result in erosion and erosion-corrosion. Experience gained from production of hydrocarbons has shown that severe degradation of production equipment may occur under erosion-corrosion conditions. Sand erosion in multiphase flows with entrained sand is a complex phenomenon. There are many issues that remain unanswered. The severity of erosion depends on a multitude of factors such as fluid properties, flow rate, sand size and rate, material type, geometry as well as many others. However, trying to isolate the effect of one of these factors and making the findings applicable to a wide range of conditions is challenging. One factor that affects the severity of erosion that has not been addressed directly in the literature, but has been observed experimentally is the pipe orientation, horizontal or vertical. The semi-mechanistic erosion prediction procedure (SPPS) was previously created by E/CRC. The SPPS extension for multiphase flows was developed based on limited data for 1 and 2 pipe diameters. Experiments have shown that pipe size and flow orientation impact the severity of erosion in annular and low liquid loading flows. Also, the current version of SPPS is for vertical orientation only. It becomes necessary to develop an extension of the current mechanistic model procedures that account for: Physics of low-liquid loading and annular multiphase flows. Effect of pipe size and flow orientation on the severity of sand erosion.

4 VIII-4 EROSION/CORROSION RESEARCH CENTER Annular flow and low liquid loading flow regimes are the most common flow patterns encountered in natural gas wellbores and pipelines. They occur under conditions of high gas flow rates and very low to medium liquid flow rates. The liquid flows as a film around the pipe inner wall, surrounding a high velocity core, which may contain entrained liquid droplets. The interface between the gas core and the liquid film is very wavy, and atomization and deposition of liquid droplets occur through this interface. Most of the experimental and theoretical studies on annular flow have been carried out either for vertical or for horizontal conditions. Changes in the physical phenomena occur as the inclination angle varies from vertical through inclined to horizontal flow conditions. Under vertical flow conditions, the time-average liquid film distribution is nearly uniform around the pipe inner periphery. As the pipe is inclined from the vertical to off vertical, the film thickness distribution becomes non-uniform. Due to gravity, the liquid phase tends to accumulate at the bottom part of the pipe. This results in a thicker film at the bottom and a thinner film at the top. The non-uniform film thickness distribution becomes more and more pronounced as the pipe inclination angle approaches horizontal. This phenomenon has a significant effect on the liquid holdup and pressure drop in the system and must be accounted for in order to enable proper design of pipelines, wellbores, and separation facilities. The following is a brief presentation of the pertinent literature for annular flow in vertical and horizontal pipes: For vertical flow, Wallis [4] and Hewitt and Hall-Taylor [5] presented general discussions of annular flow. Earlier models for annular flow were developed by Dukler [6] and Hewitt.[7] Other models have been published by Hasan and Kabir [8], Yao and Sylvester [9], Oliemans et al.[10], Caetano [11] and Alves et al.[12]. The physical mechanisms associated with annular flow have also been studied extensively. Turner et al. [13] and Ilobi and Ikoku [14] studied the minimum gas velocity required for liquid removal from vertical gas wells. Wallis [4], Henstock and Hanratty [15], Whalley and Hewitt [16] and Asali et al. [17] developed interfacial shear stress correlations. The entrainment process was studied by Hanratty and Asali [18], Schadel et al. [19] and Whalley and Hewitt [16]. For horizontal flow, measurements of the circumferential film thickness distribution were reported by McManus [20], Butterworth [21] and Fisher and Pearce [22]. Experimental data on film thickness distribution and pressure drop were acquired by Dallman [23], Laurinat [24] and Laurinat et al. [25]. In a later study, Laurinat et al. [26] developed a model for film thickness distribution. Jepson [27] evaluated the proposed model and found it suitable for use in large diameter pipes. Williams [28] conducted comprehensive studies on the effect of pipe diameter on annular flow in horizontal pipes. Annular flow in small diameter pipes was studied by Luninski et al. [29].

5 VIII-5 Previous Experimental Work on Erosion Measurements in Gas Dominant Systems The experimental approach requires using a flow geometry of interest (such as pipe, elbow, and tee) or a representative test specimen to conduct the erosion tests under specific flow conditions. The erosion ratio (mass loss of the geometry/ mass of the sand that causes erosion) and/or penetration rates (thickness loss per unit sand throughput, mils/lb or mm/kg) are then calculated from the mass loss or thickness loss data, geometry, flow and test conditions. This experimental erosion data can be used to validate erosion models. An extensive empirical information has been gathered at TUSMP for examining sensitivity of ER probes in low liquid loading multiphase flow. For example, Antezena [30] used electrical resistant (ER) probes in 1 pipes and plug tees. Pyboyina [31] conducted experiments in two different orientations, vertical and horizontal for 2 pipes. Nuguri [32] carried out experiments to examine the effect of low-liquid rates on ER probe erosion for gas dominant flows for 2 and 3 pipe diameters. Dosila [33] modified and compared with experimental data a previous erosion model that was developed for very low liquid-high gas flow. Fan [3] examined ER probes for 3 and 4 pipes under low liquid loading. Figure 1 shows ER probe results obtained by Nuguri [32] in 50.8 mm (2 ) elbows in vertical orientation, for a fixed superficial gas velocity of 30 m/s, water viscosity = 1cp. Although those experiments provide valuable information for erosivity, these ER probe data were collected in short period of time and repeatability of erosion data needs to be examined. Also, all these experiments used ER probes that are fixed at a location and distribution of erosion pattern under low liquid loading and annular flow is unknown.

6 VIII-6 EROSION/CORROSION RESEARCH CENTER 4.2E-03 ER Probe Data Erosion Ratio, mm/kg 3.6E E E E E E E Superficial Liquid Velocity, m/s Figure 1 - Effect of Superficial Liquid Velocity on ER Probe Erosion; 300 µm Sand, V SG =30 m/s, vertical orientation. Data from Nuguri [32] Sand Erosion Detection Technologies In the present work, experimental erosion measurements were performed with several techniques including electrical resistance (ER) probes and non-intrusive ultrasonic wall thickness measuring (UT) probes while varying superficial liquid and gas velocities, sand sizes and flow orientations. A CPVC test cell is also used to observe erosion pattern and location of maximum erosion on the outer wall of the elbow. Intrusive Electrical Resistence Probe The basic principle of ER probe intrusive measurement stems from electrical resistance measurements. The core element of the ER probe structure is a pair of spiral electrodes that are attached to each other, namely the sample element (or exposed element)

7 VIII-7 and reference element. The sample element is exposed to the flow, where sand particles impinge the sample element; the reference element is built inside the probe and protected from direct contact with flow. The electrical resistance of the reference element remains the same since it is protected from the flow, while the electrical resistance of sample element changes since the sand particles are causing erosion changing the cross-sectional area of the element. Then, the Cormon transmitter unit along with data acquisition software measures the real-time electrical resistance for both the sample element and reference element. The difference between the two is related to the corresponding metal loss. Non-Intrusive Ultrasonic Erosion Rate Monitoring (UT) In the present work, a non-invasive technique for measuring erosion based on ultrasonic measurements was used to collect erosion data for multiphase flows under annular conditions. The ultrasonic device used was developed by Scott Grubb, former E/CRC PhD Student and current scientist at ConocoPhilips [34]. It includes a process for measuring a thickness of a sound conducting material which involves a ultrasonic source to provide a pulse into the material and a ultrasonic receiver to collect reflections from the opposite side of the material. The temperature of the pipe is measured while a series of pulses is emitted from the source into the material. A temperature corrected wall thickness is determined based on the calculated average time for an ultrasonic sound wave to travel through the wall along with the coefficient of thermal velocity expension in a temperature compensation model. Previous Erosion Models for Gas Dominant Systems A variety of erosion prediction models are available in literature. Representatively, the Erosion Corrosion Research Center (E/CRC) at The University of Tulsa has been developing and improving the program Sand Production Pipe Saver (SPPS) which is capable of predicting the maximum sand penetration rates in single and multiphase flow for multiple oil field geometries including elbows, tees, and straight pipes. The E/CRC has also developed a Computational Fluid Dynamics (CFD) based erosion prediction procedure with which the erosion in several geometries can be calculated in single-phase flow. A review of the previous studies and efforts done to characterize erosion profile in elbow are presented. Bourgoyne [35] collected experimental data in various fittings for gas/liquid annularmist flow and provided recommendations for values of specific erosion factor in his erosion

8 VIII-8 EROSION/CORROSION RESEARCH CENTER prediction model. Salama [36] reported the data obtained by Det Norske Veritas (DNV) in elbows for a broad range of flow regimes including bubbly flow, slug flow, and annular flow. Salama also proposed an empirical correlation that accounts for mixture fluid properties of multiphase flow as an alternative of API RP 14E [37]: V e d ρm = S (VIII-1) W where V e is the erosional velocity limit; S is an empirical constant; D is pipe diameter; is the fluid mixture density; and W is sand production rate. The models proposed by Bourgoyne [35] and Salama [36] share the advantage of simplicity as well as the weakness of trying to reflect the multiphase flow effect by using a simple empirical coefficient. More recently, Mazumder [38] at E/CRC presented a mechanistic model for annular flow that was grounded on previous models that were based on mixture properties for multiphase flows. McLaury et al. [39] refined the annular flow erosion model by accounting for 2-D particle tracking through the annular film. Erosion Prediction for Annular Flow McLaury et al. [39] developed an annular flow erosion model that accounted for the flow pattern and flow orientation. In the model, the erosion is assumed to result from both particles in the core and in the liquid film and separate calculations are performed for each. Models from literature to predict liquid film thickness and entrained fraction are incorporated in the erosion calculation procedure. Based on experimental results obtained by Santos [40] in a 1 pipe, the model also assumes that sand is uniformly distributed in the liquid for sand concentration in annular flow. The model determines the erosion damage caused by particles in both regions and sums them to determine the total amount of erosion. ρ m Particles in the liquid film The particles traveling in the gas core must pass through the gas core and the liquid film before impacting the wall, so two different particle track routines are used to determine the representative impact velocity. The original form of the model [41] is applied to examine the motion of the particle through the gas core. In order to apply the original model, the following parameters must be known: stagnation length, characteristic velocity, fluid density and viscosity.

9 VIII-9 The stagnation length is calculated based on the relation for the original model using the inner diameter of the pipe. The characteristic velocity is a function of the average velocity of the gas velocity in the gas core. This is calculated [42] using the following Equation: 2 d 2 VG = VSG (VIII-2) d δ where V SG is the superficial gas velocity, d is the inner diameter of the pipe, δ is the film thickness. Particles in the gas core The initial velocity of the particle entering the liquid film is assumed to be the particle velocity calculated at the intersection between the gas core and the liquid film calculated in the previous particle tracking routine. The average film velocity is calculated [42] by: V Film = V SL 2 ( 1 E) d 4δ ( d δ ) (VIII-3) Entrainment fraction calculation Many models have been employed to predict the entrainment fraction ( E ) in sand erosion prediction models. Mazumder [38] used the correlation developed by Ishii and Mishima [43]. The entrainment liquid fraction is predicted as follows: ( We Re ) E = tanh (VIII-4) where the Weber number We and the liquid Reynolds number L Re L. An empirical correlation of liquid entrainment fraction was proposed by Oliemans et al. [44] using AERE Harwell data bank, E = 10 1 E β 0 ρ β 1 L ρ β 2 G µ β 3 L µ β 4 G σ β 5 d β 6 V β 7 SL V β 8 SG g β 9 (VIII-5) where β parameters as listed in the Table 1:

10 VIII-10 EROSION/CORROSION RESEARCH CENTER Table 1 - Summary of Parameters used in Equation (VIII-5) Parameter Physical Quantity Parameter Estimate β 0 Intercept β 1 β 2 β 3 β 4 ρ L 1.08 ρ G 0.18 µ L 0.27 µ G 0.28 β 5 σ β 6 d 1.72 β 7 β 8 V SL 0.70 V SG 1.44 β 9 g 0.46 Zhang et al. [45] reorganized this correlation as a function of six non-dimensional groups, E ρ L µ L = 0.003We Re Re 1 SG Fr SG SG SL E ρg µ G (VIII-6) where We SG, Fr SG, Re SG and Re SL are given by: We SG 2 ρg VSG d = (VIII-7) σ V Fr = SG SG g d (VIII-8) ρ V d G SG Re SG = (VIII-9) µ G ρ V d L SL Re SL = (VIII-10) µ L

11 VIII-11 After this reorganization, Zhang et al. [45] found that β 4 should be changed to 0.27 to satisfy the dimensionless requirement. E/CRC Erosion Equations In order to overcome shortcomings of previous erosion prediction models and correlations, McLaury and Shirazi [46] developed a semi-empirical procedure to predict erosion in multiphase flow. The model was an extension of a previous model that was originally developed for single-phase flow and based on Computational Flow Dynamics modeling and data [46]. The key to the model is to predict a representative particle impacting velocity, and then apply an erosion equation to quantify the erosion rate from the particle impact information. For carbon steel pipes, the form of the equation can be written as: ER 0.59 n = C( HB) F F f ( θ ) V (VIII-11) S P P where ER is the dimensionless erosion ratio which is the loss of wall material divided by the mass of particles, C is a material dependant constant, HB is the Brinell hardness, F S is the particle sharpness factor, F P is a penetration factor that depends on the material density and geometry, f ( θ ) is a function of particle impact angle that depends on the material, V P is the particle impact velocity and n is another empirical constant. The hardness function ( HB ) was developed for carbon steels. Every term in Equation (VIII-11), with the exception of the particle velocity, is empirically based. The main difference between this model and earlier models in the literature is that a representative solid particle to metal impact velocity V P is used to calculate erosion instead of the flowstream velocity. The investigators [46] developed a simplified method for calculating the characteristic particle impact velocity, which is obtained by creating a simple model of the stagnation layer representing the pipe geometry. The stagnation zone is a region that the particles must travel through to strike the pipe wall for erosion to occur. The particle velocity in this zone and resulting erosion depend on a series of factors such as fitting geometry, fluid properties, flow regimes, pipe material properties and sand properties. The initial particle velocity is set equal to the characteristic fluid velocity used at the initial particle location. A particle equation of motion is then used to determine the velocity of the particle as it approaches the wall. When the particle is a radius away from the wall, the particle tracking is stopped and the particle velocity at this location is the impact velocity.

12 VIII-12 EROSION/CORROSION RESEARCH CENTER Erosion Prediction for Low-Liquid Loading Conditions As the gas velocity is increased, intense turbulence in the gas stream may cause all the liquid to break into droplets. This type of flow is usually called dispersed liquid or mist flow. This situation can be seen as a limiting case of annular flow for a negligible film thickness. In this research, low liquid-loading conditions refers to gas dominant flows where there is insufficient liquid to form a continuous liquid film around the pipe creating annular flow. For very low liquid rates, liquid flows in the form of droplets forming liquid stripes on the pipe walls. Higher erosion rates are observed in gas dominant wells that can damage production equipment, piping and fittings. The erosion in low liquid gas flow ranges from the erosion in gas only flows to erosion in annular flow depending on the liquid rate. To develop the erosion model for low liquid gas flows, the boundaries of the flow pattern must be determined. An empirical fit was developed [33] for very low liquid-high gas flow where the superficial liquid velocity ranges from ft/s ( m/s) to the local minimum erosion near the annular flow transition. The erosion in the very low liquidhigh gas flow region is given by Equation (VIII-12): where: V sl ln V sl1 ER = (ER 2 ER1) + ER1 (VIII-12) V sl2 ln V sl1 V sl = Superficial liquid velocity of interest V sl1= Superficial liquid velocity of ft/s ( m/s) V sl2 = Superficial liquid velocity at local minimum in erosion near the transition to annular flow ER 1 = Erosion prediction for V sl1 (assuming gas only) ER = Erosion observed for V sl2 (using annular flow model) 2 In order to apply Equation (VIII-12), the value for V sl2 must be determined. Dosila [33] compared this local minimum to a ratio of particle diameter to film thickness, where the film thickness values for different flow conditions was calculated using the SPPS program. Additional comparisons were made for other low liquid gas conditions in the 2 and 3-inch loops. The results demonstrated that the local minimum in erosion occurs at film thickness

13 VIII-13 to particle diameter ratios between 0.8 and 2.0. Other results showed that the predicted trends of erosion obtained from SPPS program using a ratio of film thickness to particle diameter value of 1 are also relatively close to the experimental normalized erosion obtained from ER probes for other flow conditions. Sand Erosion Measurements in Elbows To further investigate the applicability of the models developed to predict erosion rates under high gas and low liquid rates, multiple experiments has been conducted in three different multiphase flow loops at different gas, liquid and sand rates. These facilities were designed and constructed by keeping sufficient upstream length to promote fully developed flow near the erosion test section. All the experiments in this facilities were conducted by keeping the flow loop in the horizontal and vertical positions. Also, it should be noted that some of the data obtained at TUSMP were aimed at examining the sensitivity of intruments and erosion data were obtained during a short period of time were large uncertainity may exist on the absolute values of data that was collected. At present, only three and four-inch test sections are being used to conduct erosion experiments at E/CRC. One-Inch Flow Loop Mazumder [38] used a 1 test facility to measure local thickness loss measurements in elbows specimens. In this test facility, sand and liquid were mixed in a small slurry tank. Then sand and liquid were injected into the gas stream. The gas flow velocity in this oneinch test section can reach to about 100 ft/s (30 m/s) at a pressure of about 40 psig. The gasliquid-sand mixture then flows through the test section. After the test section, the mixture flows to a cyclone separator where liquid and sand are separated from the gas stream. The liquid and sand are discharged and the gas flows back to the existing gas-liquid separator before it flows back to the compressor. These test sections uses test cells with an erosion specimen. The test cell containing the test specimen was made of two halves of CPVC. A 90 elbow specimen of ¼ x ¼ cross-section was placed in the test cell that simulates the outer wall of a 1 inch elbow. Two-Inch Flow Loop The two-inch Flow Loop was constructed to perform tests in different flow regimes. The two inch flow loop consists of two 100-scfm compressors, one 90 gpm (340 liters/min) liquid pump, a 1000 gallon (3785 liters) liquid tank, gas and liquid flow meters,

14 VIII-14 EROSION/CORROSION RESEARCH CENTER approximately 90 ft (27 m) section of 2 inch (50.8 mm) pipe. A section of the 2-inch (50.8 mm) pipe is constructed using Plexiglas for flow visualization. The two-inch flow loop is capable of superficial liquid velocity of up to about 10 ft/sec (3 m/s) and a superficial gas velocity of up to approximately 100 ft/sec (30 m/s). This facility was used to develop the existing mechanistic prediction model through many experimental studies using ER probes and acoustic monitors. Large Scale Boom Loop The large scale flow loop has a basic working procedure similar to the two-inch flow loop. Schematic views of the testing facilities are shown in Figures 2 and 3. Figure 2 - Schematic of Large Scale Boom Loop

15 VIII-15 Figure 3 - Test Section of Large Scale Boom Loop The main difference between the small loops and the large boom loop is that more operating and control devices were added to achieve a broader range of conditions and allowing higher gas velocities. Two-inch, three inch and four-inch sections can be mounted on the large boom loop which can be positioned at any angle. Current Experimental Results Experimental Results Using ER Probes Experimental erosion studies on flat-head ER probes at 45 in the elbow were conducted by on the Large Boom Loop in two different orientations, horizontal and vertical for 3 inch (50.8 mm) pipe diameter. Experiments were performed for different flow regimes ranging from gas-sand only conditons to churn-annular transition. The results of ER probe measurements for a superficial gas velocity of 27 m/s are shown in Figure 4. From Figure 4, it can be seen that considerable change in metal loss rate is observed when the velocity of liquid changes under different flow conditions. With a further increase in the amount of liquid added, a transition to an intermediate regime between low-liquid gas flow and annular flow starts. Then, for vertical flow orientations, erosion reaches a local minimum value as the liquid velocity approaches annular flow. These are similar to results obtained by Nuguri in Figure 1 for 2 pipe.

16 VIII-16 EROSION/CORROSION RESEARCH CENTER Figure 4 Effect of Superficial Liquid Velocity on ER Probe Measurements Erosion; V SG = 27 m/s; 300 µm Sand Size; 1 cp water viscosity; 76.2 mm (3 ) Elbow Experimental erosion studies on flat-head ER probes at 45 in the elbow were conducted by on the Large Boom Loop in vertical orientation for 76.2 mm elbows for low gas velocity. The observed flow pattern was annular with flow reversal. A schematic of the flow structure of this flow pattern is shown in Figure 5. Recent experimental studies at Tulsa University Fluid Flow Projects has shown that for lower gas velocities, the liquid film has a complex flow pattern as shown schematically in Figure 5. The results of ER probe measurements at 45 for vertical orientations using 300 µm Sand for a superficial gas velocity of 27 m/s are shown in Figure 6 that show erosion rates decrease significantly as liquid rate increases in the pipe. :

17 VIII-17 Figure 5 - Schematic cross-section of Annular with Flow Reversal (V SG = 27 m/s; V SL = 0.02 m/s) Figure 6 Effect of Superficial Liquid Velocity on ER Probe Measurements Erosion; V SG = 15 m/s (Vertical annular with flow reversal); 300 µm Sand Size; 1 cp water viscosity; 76.2 mm (3 ) Elbow In this research period, additional ER-Probe measurements were conducted using flat-head ER probes at 45 in the elbow on the Large Boom Loop in two different orientations, horizontal and vertical for 4 pipe diameter. Experimental condition are shown in Table 2. The results of metal loss rate for both flow orientations are shown in Figure 7.

18 VIII-18 EROSION/CORROSION RESEARCH CENTER V SG (m/s) Table 2 Experimental Condition for ER-Probe Measurements V SL (m/s) Horizontal vs. Vertical - Air Water Annular Flow 4 Pipe Sand Size (μm) Water Viscosity (cp) Sand Throughput (g) Water Throughput (gal) Sand Concentration (%) Erosion Rate, mm/yr VERTICAL HORIZONTAL Flow Orientation Figure 7 - Comparison Horizontal vs. Vertical - Air Water Annular Flow 4 Pipe It can be seen that at a superficial gas velocity of 30.5 m/s for the vertical flow orientation, the metal loss rate is approximately 1.5 times higher than the metal loss rate for the horizontal flow orientation at a superficial liquid velocity of 0.19 m/s. The ratio of erosion magnitudes between vertical and horizontal is not as severe as the low liquid cases [33]. The possible reasons for this behavior may be as follow: The particle impact velocities may be lower in the horizontal case than the vertical case There may be higher sand entrainment into the gas core region in vertical pipe orientation

19 VIII-19 Experimental Results Using Ultrasonic Transducers In the present work, UT measurements were taken on a 3 ID (76.2mm) Stainless Steel elbow on the Large Boom Loop in two different orientations, vertical and horizontal. Vertical Orientation Data from a total of seventeen annular flow experiments will be displayed and processed to show the detection of the vertical annular flow regime erosion pattern. Numbering of the mounted ultrasonic transducers on the elbow for vertical orientation is shown in Figure 8. Experiments were conducted with superficial gas velocities ranged from 15 m/s to 49 m/s and superficial liquid velocities ranged between m/s and 0.5 m/s. Sand concentration was held constant for all the experiments at 1% by weight. Sand size was 150 µm for three of the experiments and 300 µm for the rest. Viscosity of the water was 1 cp for all the experiments. Figure 8 Transducers Number and Location in the Elbow (Vertical Orientation) Maximum measured erosion rates ranged from 8 mm/yr to 667 mm/yr. Experimental conditions, erosion rate and erosion ratios for vertical conditions are shown in Table 3.

20 VIII-20 EROSION/CORROSION RESEARCH CENTER Table 3 Experimental Conditions and Erosion Results Using Ultrasonic Transducers for Vertical Gas-Liquid Flows in 76.2 mm (3 ) Stainless Steel Elbow Figure 9 shows a sample wall loss rate results obtained for V SG =49 m/s and V SL =0.2 m/s. Units of measurement are mm per year (mm/yr). Figure 9 Sample UT Erosion Rate Measurements in mm/yr for Vertical Annular flow, V SG =49 m/s and V SL =0.2 m/s.

21 VIII-21 Using all the data generated from the seventeen erosion experiments in vertical conditions, the averages and the 95% confidence intervals of the percent of maximum erosion values were calculated for each individual transducer location. Figure 10 shows these group average percent of maximum erosion values along with their associated 95% confidence intervals at each transducer location. Sixteen of the seventeen annular flow experiments had its maximum erosion measured at the same transducer location, which was position number 7. Figure 10 Percent of Maximum Erosion Average and 95% Confidence Interval. Vertical Annular Flow. Units in %. From Figures 9 and 10 it can be seen that for vertical flow orientation, the maximum erosion values obtained were localized approximately in the middle plane section at 45 in the elbow. Also, nearly symmetrical erosion pattern is observed from wall loss rate in vertical flows. Figure 11 shows the comparison of two erosion experiments. The magnitudes of erosion measured by all the transducers were considered for the comparison. From Figure 11, it can be concluded that the location of maximum erosion is similar for the experiments and the experimental magnitudes are comparable. The data show that, except for a few locations, the magnitude of erosion rates are nearly similar for both experiments.

22 VIII-22 EROSION/CORROSION RESEARCH CENTER Figure 11 - Comparing Magnitudes of Erosion Measured by Transducers V SG =49 m/s, V SL =0.02 m/s, 300 μm Sand, Water 1 cp Viscosity, 1 % conc. by weight Horizontal Orientation Data from a total of eleven annular flow experiments are displayed and processed to show the detection of the vertical annular flow regime erosion pattern. Numbering of the mounted ultrasonic transducers on the elbow for vertical orientation is shown in Figure 12. Figure 12 Transducers Number and Location in the Elbow (Horizontal Orientation)

23 VIII-23 Erosion data has been obtained for superficial gas velocities ranged from 35 m/s to 49 m/s and superficial liquid velocities ranged between 0.5 m/s and m/s. Sand concentration was held constant for all the experiments at 1% by weight. Sand size and liquid viscosity was 300 µm and 1 cp, respectively, for all the experiments. Maximum measured erosion rates ranged from 15 mm/yr to 89 mm/yr. Experimental conditions, erosion rate and erosion ratios for vertical conditions are shown in Table 4: Table 4 Experimental Conditions and Erosion Results Using Ultrasonic Transducers for Vertical Gas-Liquid Flows in 76.2 mm (3 ) Stainless Steel Elbow A sample erosion pattern and data in mm per year (mm/yr.) is observed in Figure 13 that shows wall loss rate results obtained for V SG =49 m/s and V SL =0.04 m/s. Figure 13 - UT Erosion Rate Measurements in mm/yr. Horizontal Annular flow, V SG =49 m/s and V SL =0.04 m/s

24 VIII-24 EROSION/CORROSION RESEARCH CENTER Using all the data generated from the eleven erosion experiments in horizontal conditions, the averages and the 95% confidence intervals of the percent of maximum erosion values were calculated for each individual transducer location. Figure 14 shows these group average percent of maximum erosion values along with their associated 95% confidence intervals at each transducer location. Nine of the eleven annular flow experiments had its maximum erosion measured at the same transducer location, which was position number 7 (42 location in the outer section of the bend): Figure 14 Percent of Maximum Erosion Average and 95% Confidence Interval. Horizontal Annular Flow. Units in %. From Figures 13 and 14, it can be seen that for horizontal flow orientation, the maximum erosion values obtained were localized approximately in the middle plane section at 45 in the elbow. Also, an asymmetrical erosion pattern is observed from wall loss rate in horizontal flows. More erosion is observed on the upper section of the elbow. The possible reasons for this behavior may be the liquid film distribution in horizontal flows. Figure 15 shows the comparison of two erosion experiments. The magnitudes of erosion measured by all the transducers were considered for the comparison. From Figure 15, it can be concluded that the location of maximum erosion is similar for the experiments and the experimental magnitudes are comparable.

25 VIII-25 Figure 15 - Comparing Magnitudes of Erosion Measured by Transducers V SG =40 m/s, V SL =0.02 m/s, 300 μm Sand, Water 1 cp Viscosity, 1 % conc. by weight CPVC Test Cell Results An erosion test cell with the elbow specimen was used in this research to observe paint erosion patterns and measure metal loss in an aluminum specimen. Figure 16 shows the test cell configuration, flow directions and the location of the specimen inside the elbow: Figure 16 CPVC Test Cell

26 VIII-26 EROSION/CORROSION RESEARCH CENTER The test cell is made of two halves of CPVC. A 90 elbow specimen of ¼ inch by ¼ inch (6.35 mm by 6.35mm) cross-sectional area is placed inside the test cell that simulates the outer wall of a 3 inch (76.2 mm) elbow with ratio of 1.5. Experimental studies have been performed to demonstrate the possibility of using a paint erosion approach to accurately identify erosion patters. For the paint method, inside of the CPVC elbow is painted. When sand is injected, paint can be removed from the locations which the sand is impacting and one can directly observed these locations. Table 5 shows experimental condition for erosion measurements. Figure 17 shows the paint erosion pattern obtained after 10 minutes of experiment for vertical annular flow: Flow Orientation Table 5 Experimental Conditions for Test Cell Air Water Annular Flow 3 in (76.2 mm) Elbow. 1cP Water Viscosity V SG (m/s) V SL (m/s) Sand Size (μm) Sand Throughput (g) Water Throughput (gal) Sand Concentration (%) Vertical Horizontal Figure 17 Paint Erosion Test Result, Vertical Annular Flow, V SG = 49 m/s, V SL = 0.2 m/s, water 1 cp, 1 % concentration by weight, 300μm Experiment Time = 10 min.

27 VIII-27 Figure 18 shows the paint erosion pattern obtained after 30 minutes of experiment for horizontal annular flow. Figure 19 illustrates the vertical specimen thickness loss profile for V SG = 49 m/s, V SL = 0.2 m/s. From Figures 18 and 19, it can be observed that different paint erosion patterns are present for each flow orientation condition in annular flow. So it confirms the results obtained using ultrasonic transducer for similar operating conditions. From Figure 19, it can be observed that the maximum thickness loss of 148 microns was measured at approximately 45 degrees from the inlet of the elbow. So, the location of maximum erosion location agrees with the results obtained using the non-intrusive technique. : Figure 18 Paint Erosion Test Result, Horizontal Annular Flow, V SG = 49 m/s, V SL = 0.2 m/s, water 1 cp, 1 % concentration by weight, 300μm Experiment Time = 30 min.

28 VIII-28 EROSION/CORROSION RESEARCH CENTER Figure 19 Thickness Loss Profile of Elbow Aluminum Specimen, V SG = 49 m/s, V SL = 0.2 m/s, water 1 cp, 1 % concentration by weight, 300μm Experiment Time = 10 min. Recent Improvements of Erosion Models The erosion model for annular flow and low liquid loading was developed based on earlier data for smaller pipes and also based on data that was obtained at TUSMP for evaluating ER probes. Recent experimental work [2] on two-phase vertical upward flow has shown that pipe diameter has an effect on many flow structure parameters in annular flows such as film thickness and entrainment rate. Thus, it becomes necessary to carefully evaluate multiphase flow models that are incorporated in SPPS. The model development is based on many components and in the present work all components of the model are being evaluated. Figure 20 shows a flowchart that describes all the various components that the erosion model is based upon.

29 VIII-29 Figure 20 - Model Development Flowchart for Erosion in Low-Liquid Loading and Annular Flows The specifications of model development for sand erosion prediction under lowliquid conditions and annular flows include: a) Selection and validation of mechanistic models for entrainment fraction and film thickness calculation in both vertical and horizontal flows. b) Estimation of sand rate and particle velocities that are responsible for erosion. c) Measure sand distribution and sand concentration in low-liquid and annular flows to study how sand distribution can affect erosion results under both horizontal and vertical orientations. d) Evaluation and improvement of equations that are used to calculate the equivalent flowstream velocity; computation of the characteristic impact velocity of the particles through particle tracking routines. e) Evaluation and validation of new erosion equations developed to predict erosion rate for different materials used in oilfield industry. f) Improvement of flow field estimation in elbows geometries and calculation of representative particle trajectories that are used as the impingement information in erosion equations to determine a penetration rate. g) Using new erosion measurements in elbows geometries, validate the model.

30 VIII-30 EROSION/CORROSION RESEARCH CENTER h) If the model does agree with erosion data, it can be include in SPPS modules. If not, the results can be used as feedback for future improvements. Different actions related to some of these specific steps explained in the flowchart are presented below. Validation of Mechanistic Model for Gas-Liquid Annular Flows Entrainment Fraction Validation Magrini [47] carried out 140 experiments on a 76.2 mm (3 in) internal diameter pipe at TU-Fluid Flow Project facilities in North Campus. Air and tap water were used as working fluids. Experimental techniques used in that study include iso-kinetic sampling probes, film removal devices, conductivity film thickness measurements, and quick-closing valves. In the present work, the experimental results obtained from that study were used to evaluate existing methods for predicting entrainment fraction. Figure 21 displays the results of the comparisons between entrainment fraction data in vertical annular flow and the predicted entrainment fraction obtained from SPPS 4.2 using the entrainment fraction correlation developed by Zhang et al. [45]. As it can be seen in Figure 21, the predicted entrainment rates using Zhang s model (Equation VIII-5) slightly overpredict entrainment for all gas velocities. 1 Perfect Agreement SPPS 4.2 Predicted Entrainment Fraction [-] Entrainment Fraction Data [-] V SL = m/s V SG = m/s Figure 21 - Comparison of Entrainment Model Prediction with Experimental Entrainment Fraction Data. Data from Magrini [47]

31 VIII-31 The entrainment fraction is used in the mechanistic model to determine the fraction of sand particles in the gas and in the liquid film. Since prediction of entrainment rate is important in development and improvement of the current mechanistic model, further experimental entrainment fraction data is required on two-phase vertical flow in 3 and 4 diameter pipes. Film Thickness Calculation in Vertical Annular Flow Kaji and Azzopardi [48] correlated the film thickness with liquid film Reynolds number as in the following Equation: δ = (VIII-13) + B L ARe LF where A and B are constants, ρ V d L F F Re LF = (VIII-14) µ L d F ( d ) δ L δ L = 4 (VIII-15) d + δρ L δ L = µ L τ i ρ L (VIII-16) where Re LF is the Reynolds number of the liquid film, d F is the hydraulic diameter of the liquid film, δ is film thickness. The interfacial shear stress τ i is given by: 1 2 τ i = fiρgvsg (VIII-17) 2 where f i is interfacial friction factor. Asali [49] proposed a correlation for the interfacial friction factor in annular flow. The correlation was modified by Ambrosini [50] as: f f i G = We D Re 0.6 G + δ L 200 ρg ρ L (VIII-18) where the Weber number We G, and the Reynolds number Re G are defined as:

32 VIII-32 EROSION/CORROSION RESEARCH CENTER We G 2 ρg VC d = (VIII-19) σ ρ V d G C Re G = (VIII-20) µ G The film velocity V F and gas core velocity V C is given, respectively, by: V F = V SL 2 ( 1 E) d δ ( d δ ) 4 L L (VIII-21) V C ( VSG + VSL ) ( d δ ) 2 2 E d = (VIII-22) 2 L The dimensionless thickness of the liquid film is: + δρg δ G = µ G τ i ρ G (VIII-23) and the single phase friction factor is: f (VIII-24) 0.2 = 0.046Re G G Film Thickness Model Validation using Experimental Data Since the liquid films involved in most types of film flows are rather thin ( < 3mm) accurate measurement of their thickness is not easy and many alternative methods for their measurement and calculation have been proposed. The methods can be classified into three main groups: film average methods, localized methods and point methods. Magrini [47] used conductance probes to measure liquid film thickness in air-water annular flow on a 76.2 mm internal diameter pipe. The data was collected for superficial gas velocity of m/s and superficial liquid velocities of m/s and 0.04 m/s. Van der Meulen [51] also used conductance probes in annular gas-liquid in a 127 mm (5 in) diameter vertical pipe. The data was collected for superficial gas velocities from 4.5 m/s (15 ft/s) to 23 m/s (75 ft/s) and superficial liquid velocities from m/s ( ft/s) to 0.3 m/s (0.98 ft/s). The predictions were compared with both groups of experimental data for the purpose of their validation.

33 VIII-33 Figure 22 - Comparison of Film Thickness Model Predictions with Experimental Data at V SL =0.49 ft/s. Data from Van der Meulen [51] Predicted Film Thickness [mm] Perfect Agreement SPPS Film Thickness Data [mm] V SL = m/s V SG = m/s Figure 23 - Comparison of Film Thickness Model Predictions with Experimental Data Data from Magrini [47]

34 VIII-34 EROSION/CORROSION RESEARCH CENTER As it can be seen in Figures 22 and 23, the correlation sometimes overestimates film thickness for a 76.2 mm diameter pipes and slightly underestimates film thickness for a larger pipe diameter. In the present work, the correlation used by Kaji and Azzopardi [48] has been included in SPPS 4.2 calculations. New film thickness data for both small and large diameter pipes available in literature will be used to validate and improve film thickness correlations included in SPPS. Sensitivity Analysis of Film Thickness and Entrainment Fraction in Erosion Calculations In order to examine relationships between annular flow characteristics and erosion predictions models, a sensitivity analysis (SA) is presented. The values of entrainment fraction and film thickness in vertical gas-liquid flows were changed in the current erosion prediction model to determine the effects of such changes in sand erosions calculations. An experimental condition applied in ER probe erosion tests for 4 inch (101.6 mm) bends was used. The parameters are presented in Table 6. As it can be seen in Figure 24, as the value of film thickness decreases 50%, the erosion prediction values increase 40% approximately. On the other hand, if the film thickness increases 100%, the erosion prediction decreases 35% approximately. On the other hand, from Figure 25, it is observed that if the entrainment fraction decreases 75%, the erosion prediction values also decreases 75% approximately. On the other hand, if the entrainment fraction increases 50%, the erosion prediction also increases 50% approximately. This is because it is assumed that the amount of sand that is flowing in the gas core is proportional to the amount of liquid that is entrained in the gas phase. In summary, from the presented results it can be said that the erosion prediction model is more sensitive to changes in the entrainment fraction than the film thickness calculation. However, both parameters are related and they are both important. Dia [in] Table 6 Physical conditions and Flow Parameters for Erosion Sensitivity Analysis V SG [ft/s] V SL [ft/s] ρ G [kg/m 3 ] ρ G [kg/m 3 ] µ G [cp] µ L [cp] σ [dyne/cm] SDia [μm] ρ SAND [kg/m 3 ] Sand Conc. [kg/kg] %

35 VIII E-03 DATA 0.5δ 1.0δ 1.75δ 2.0δ 2.5E-03 Erosion, [mils/lb] 2.0E E E E E Superficial Liquid Velocity, Vsl [ft/s] Figure 24 - Vertical Annular Erosion Model Sensitivity to Film Thickness (δ) Dia=4 3.0E-03 DATA 0.25E 0.75E 1.0E 1.5E 2.5E-03 Erosion, [mils/lb] 2.0E E E E E Superficial Liquid Velocity, Vsl [ft/s] Figure 25 - Vertical Annular Erosion Model Sensitivity to Entrainment Fraction (E) Pipe Diameter=4

36 VIII-36 EROSION/CORROSION RESEARCH CENTER Comparison of Erosion Data with current SPPS Model The modifications made in the mechanistic model for gas-liquid annular flows were included in the previous version of SPPS. New erosion results for ER flat-head probes at 45 in the elbow were compared with the current verison of SPPS. The result are shown in Figure 26: Predicted Erosion Ratio, mm/kg 1.0E E E E E E-06 SPPS 4.3 Perfect Agreement 1.0E E E E E E-01 Experimental Data, mm/kg Figure 26 - Comparison of ER Probe Data with SPPS 4.3 V SG = m/s, V SL = m/s, Water 1 cp and 10 cp, 1 % concentration by weight, Particle Sizes = 300 μm and 150 μm On the other hand, new erosion results obtained using Ultrasonic Transducer were compared with the current verison of SPPS. The result are shown in Figure 27. The comparison of new erosion data with current SPPS 1D version predictions shows that both E/CRC Inconel and Stainless Steel models overpredict erosion results.

37 VIII-37 SPPS 4.3 Perfect Agreement Predicted Erosion Ratio, mm/kg 1.0E E E E E E E E E E-01 Experimental Data, mm/kg Figure 27 - Comparison of UT Data with SPPS 4.3 V SG = m/s, V SL = m/s, Water 1 cp, 1 % concentration by weight, 76.2 mm (3 ) Elbows, Particle Sizes = 300μm and 150μm New Modeling Improvement Plans In order to improve the reliability of model predictions for sand erosion prediction under low-liquid conditions and annular flows new modeling improvement plan has been considered in the present work, which includes: a) Experimental investigation of liquid and gas phase velocities, b) Estimation of film thickness magnitudes in elbows, c) Develop a new erosion equation for wet surfaces

38 VIII-38 EROSION/CORROSION RESEARCH CENTER Figure 28 New Modeling Improvement Flowchart for Erosion in Low-Liquid Loading and Annular Flows Experimental Investigation of Gas and Liquid Phase Velocities In this study, a wire-mesh technique based on conductance measurements was applied to investigate two-phase horizontal pipe flow. The instrumentation and raw data extraction software was developed by Helmholtz-Zentrum Dresden-Rossendorf (HZDR- Germany). Figure 29 shows a schematic diagram of the sensor used in this study. Figure 29 - Schematic representation of a wire-mesh sensor.

39 VIII-39 The horizontal flow test section consisting of a 76 mm I.D pipe, 18 m long, was employed to generate annular flow conditions. Figure 30 shows the experimental conditions used for wire-mesh sensor (WMS) measurements: 10 1 Stratified Annular Stratified Wavy Operating Points VSL [m/s] V SG [m/s] Figure 30 - Flow Pattern of Experimental Conditions Used to Collect Data in Horizontal Gas-Liquid Flow for 1cP Liquid Viscosity. Figure 30 shows the raw time-series void fraction data for horizontal annular flow V SG = 33.5 m/s, V SL = 0.2 m/s, water 1 cp viscosity; Figure 31 - Time Series of Void Fraction. Horizontal Annular Flow. V SG = 33.5 m/s, V SL = 0.2 m/s, water 1 cp.

40 VIII-40 EROSION/CORROSION RESEARCH CENTER From the obtained raw data time series of void fraction, mean void fraction and characteristic liquid film velocities will be determined for different liquid and gas superficial velocities that ranged from 0.03 to 0.2 m/s and from 9.1 to 40 m/s, respectively. Comparisons with results obtained from existing modeling approaches available in literature for stratified and annular gas-liquid flows will be presented in future developments. Film Thickness Measurements in Bends Abdulkadir [52] carry out experimental investigations in a large diameter pipe of 127 mm attached to a vertical 180 return bend having a 381 mm radius of curvature rig using air-water as the model fluids. The objectives were to measure film fraction and local liquid film distributions in the annular flow and churn-annular flows using conductance ring probes and electrical conductance techniques. The 180 return bend was made by bolting together two slabs of transparent acrylic resin (Perspex) in the surface of each circular groove with an accurate semi-circular cross-section had been machined. The bend is of a modular construction and a probe can be inserted at radial positions of 45, 90 or 135 around the bend as shown in Figure 32: Figure Return Bend and Locations of the Conductance Probes. From experiments results, Abdulkadir noticed that for low liquid and higher gas flow rates, due to the action of gravity drainage, film breakdown occurs at the 45 bend" [52]. For low liquid flow rates and high gas superficial velocities, film break down (burn out) occurs at the 45 position around the bend. The burn out phenomenon was clearly the result of total loss of liquid from the liquid film by evaporation and entrainment. This is

41 VIII-41 confirmed by the liquid film thickness measurement. Figure 33 shows the variation of average film thickness from film fraction measurements with superficial liquid velocity around the 180 return bend at a fixed superficial gas velocity of 15.5 m/s using air-water as fluids: Film Thickness, mm degrees 90 degrees 17D UPSTREAM Superficial Liquid Velocity, m/s Figure 33 Film Thickness Data in Bends. V SG 15.5 m/s, 127 mm Pipe Diameter, Air-Water Vertical Annular Flow For the bend pipe section, at the higher gas superficial velocities, the minimum average film fractions are observed 45 bend locations. It is interesting to observe that the average film fraction for the 45 at gas superficial velocity of 10.5 to 15 m/s and liquid superficial velocity of 0.02 m/s is almost zero, suggesting that there is a film breakdown within the vicinity [52]. Film Thickness Prediction in Annular Two-Phase Flow through Pipes and Elbows using CFD A preliminary numerical calculation has been made to examine CFD capability in predicting various parameters. The presented simulation for annular gas-liquid flow focused in the reconstruction of the gas-liquid interface in the elbow. This was achieved by using VOF (Volume of Fluid) model included in the commercial CFD package STAR-CD. In this model, a single set of momentum equation is shared by the liquid film and the gas core. The

42 VIII-42 EROSION/CORROSION RESEARCH CENTER volume fraction of all the fluids in each computational cell is tracked throughout the domain. The gas-liquid interface is tracked based on the distribution of the volume fraction of the liquid in the computational cells. The flow domains were two-dimensional. The minimum cell size was Δx = 0.1mm. A scheme of the mesh made up by Cells and faces is depicted if Figure 34: Figure 34 - Schematic Representation of the Geometry and Details of Computational Grid. The simulations were implicit unsteady with a time step set at 0.5 s. The Realizable K-Epsilon Two Layer turbulence model was used. The VOF High Resolution Interface Capturing (HRIC) with 2nd Order Discretization scheme was applied for the convective terms in the volume fraction equation. The simulation time was 200 s. All physical and geometrical parameters were determined by annular air-water experimental set-up used by Van Der Meulen [51]: Experimental conditions: Pipe diameter = 127mm 5 V SL =0.301 ft/s, V SG =19.51 ft/s Pressure = 2 Bar Liquid Inlet Boundary Experimental Film Thickness δ = 1.92 mm with E= Liquid film velocity = V L = m / s Inlet boundary and gas properties Liquid film velocity = V G = 20.88m / s 3 Gas Core properties = ρ 8.23kg / m, 5 µ = Pa s C = C

43 VIII-43 Outlet: average static pressure (2 Bar) Air-Liquid Interaction (Surface Tension σ = 0.073N/m) 3.0 CFD VOF Prediction Calculated Experimental Film Thickness, [mm] Axial distance, [mm] Figure 36 Axial Development of the Liquid Film Thickness in Vertical Pipe From Figure 36, it is observed that as the present model does not consider liquid entrainment in the gas core, the VOF predictions slightly overpredict both calculate and experimental values in the straight section of the pipe. Figure 37 shows the film thickness distribution at the outer wall of the bend. Film Thickness, (mm) CFD VOF Prediction Angle, (degrees) δ Figure 37 - Film Thickness Distribution at the Outer Wall of the Bend